Downloaded from genesdev.cshlp.org on September 26, 2021 - Published by Cold Spring Harbor Laboratory Press REVIEW SR proteins and splicing control James L. Manley and Roland Tacke Department of Biological Sciences, Columbia University, New York, New York 10027 USA Twenty years have passed since the discovery of pre- ing on how they might participate in splicing regulation. mRNA sphcing (for review, see Sharp 1994). The studies A related review, including discussion of what appears to leading to this discovery were carried out on the adeno­ be an extended superfamily of SR-type proteins, has been virus late transcript, which undergoes complex alterna­ published recently (Fu 1995). tive splicing, and, therefore, the concept of alternative splicing is also 20 years old. In the intervening time, intervening sequences, or introns, have been found in a ASF/SF2, SC35, and the SR protein family vast majority of higher eukaryotic genes, and a large frac­ The first two SR proteins were discovered as a result of tion of intron-containing transcripts have been shown to biochemical studies of mammalian splicing. The proto­ be subject to alternative splicing. Furthermore, modula­ typical SR protein, ASF/SF2 (also referred to as SF2/ tion of pre-mRNA splicing is now known to be a wide­ ASF), was discovered and characterized independently spread mechanism of gene control. But despite consider­ by two different groups employing two different assays. able advances in our understanding of the splicing reac­ ASF was purified as an activity that could influence se­ tion per se (for review, see Moore et al. 1993; Madhani lection of alternative 5' splice sites in an SV40 early pre- and Guthrie 1994), insights into how alternative splicing mRNA (Ge and Manley 1990), suggesting a role for ASF is controlled have been slower to emerge. in alternative splicing, and recapitulating in vitro a pre­ Several factors have been responsible for our some­ viously observed cell-specific pattern of alternative splic­ what limited understanding of how splicing can be reg­ ing (Fu and Manley 1987). In contrast, SF2 was identified ulated. First, the combined use of biochemical and ge­ initially as an activity required for splicing of a ^-globin netic methodologies available in yeast has greatly accel­ pre-mRNA in vitro (Krainer and Maniatis 1985) and was erated understanding of the basic splicing reaction. But subsequently purified as a factor essential for splicing true alternative splicing (i.e., selection of alternative (Krainer et al. 1990a), but with activities in alternative splice sites) does not appear to occur in Saccharomyces splicing related to those of ASF (Krainer et al. 1990b). ceievisiae, and studies on this process are therefore lim­ Isolation of cDNAs encoding the proteins confirmed that ited to higher eukaryotes. With the important exception they are identical (Ge et al. 1991; Krainer et al. 1991). of the Diosophila sex determination pathway (for re­ These early studies established an important point that view, see Baker 1989), what we know about alternative remains central to our understanding of SR protein func­ splicing comes almost entirely from in vitro biochemical tion: The proteins can function in vitro both as essential approaches without the aid of genetic approaches. Sec­ splicing factors, required for splicing of all pre-mRNAs ond, again with a limited number of exceptions (for re­ tested to date, and as modulators of alternative splicing. view, see Rio 1993), the identification of cis-acting se­ The sequence of ASF/SF2 immediately suggested that quences responsible for modulating specific splicing the protein might function in vivo in splicing control events has been difficult, and, therefore, the isolation of and, furthermore, could be a member of an evolution- sequence-specific RNA-binding splicing factors has been arily conserved family of splicing factors. ASF/SF2 was slow. Indeed, whether gene-specific regulators play a ma­ found to share two significant features in common with jor role in splicing control, as sequence-specific DNA- genetically defined regulators of splicing in Diosophila. binding proteins do in transcription, constitutes an im­ First, the protein contains an amino-terminal RNA rec­ portant, unresolved question in the field. Finally, the ognition motif, or RNP-type RNA-binding domain number of regulated splicing events that have proven (RBD). Although such domains had been observed in amenable to in vitro analysis is small, adding to the dif­ splicing factors, they were also known to be present in a ficulties in the identification of regulatory molecules wide variety of RNA-binding proteins not involved in and mechanisms. Despite these limitations, significant splicing (for review, see Birney et al. 1993; Burd and progress has been made in the identification and analysis Dreyfuss 1994). A carboxy-terminal region enriched in of a family of proteins, the SR proteins, that likely play repeating arginine-serine dipeptides (the RS domain), important roles in splicing control. (The name SR re­ however, was more strongly suggestive of a role in splic­ flects the presence of a characteristic serine/arginine- ing. (For historical reasons, we refer to RS domains and rich domain; see Fig. I.) The purpose of this review is to SR proteins, even though both RS and SR refer to the summarize our knowledge of these proteins, concentrat­ same arginine-serine dipeptides.) Similar motifs had GENES & DEVELOPMENT 10:1569-1579 © 1996 by Cold Spring Harbor Laboratory Press ISSN 0890-9369/96 $5.00 1569 Downloaded from genesdev.cshlp.org on September 26, 2021 - Published by Cold Spring Harbor Laboratory Press Manley and Tacke Figure 1. Schematic representation of the domain organization of human SR proteins. SRp20 All SR proteins contain an amino-terminal RBD (red) and a carboxy-terminal RS do­ main (green). In five of the known SR pro- SC35 teins the RS domain is immediately pre­ ceded by an additional degenerate RBD (blue). The RBDs of individual SR proteins 9G8 display significant sequence similarities to the corresponding domains in other SR pro­ teins but not to each other. The highest de- SRp30c gree of sequence divergence is found in the portions of the molecules (orange) separat­ ing the first and second RBD, or the RBD ASF/SF2 and the RS domain when only one RBD is present. ASF/SF2 (Ge et al. 1991; Krainer et al. 1991) contains a stretch of nine consec- SRp40 utive glycine residues (G) in this region, whereas the corresponding part of SRp40 (Screaton et al. 1995) is composed of 30% SRp55 arginine residues. The predominant resi­ dues present within these regions in other SR proteins are glycine (G), arginine (R), pro- SRp75 line (P), and/or hydroxylated amino acids. In addition, 9G8 contains a putative zinc knuckle of the CCHC family |Z), previously found in other nucleic acid binding proteins (Cavaloc et al. 1994). Note that SC35 (Fu and Maniatis 1992) was cloned as PR264 in human and chicken (Vellard et al. 1992). Human and chicken PR264 are 98% identical. SRp20 (Zahler et al. 1992) was origmally cloned m mouse as XI6 (Ayane et al. 1991), to which it is identical, and SRp40 as HRS in rat (Diamond et al. 1993). SRp55 and SRp75 were cloned, respectively, by Screaton et al. (1995) and Zahler et al. (1993b). Among Drosophila SR proteins, RBPI (Kim et al. 1992) is the apparent homolog of SRp20, although it displays a similar degree of identity (-65%) to 9G8, whereas B52 (Champlin et al. 1991), which is likely Drosophila SRp55 (Roth et al. 1991), shares -65% sequence identity with human SRp55. been observed previously in only a few proteins, all from 20 to 75 kD. Proteins of similar sizes could be iso­ known to be involved in splicing, with the closest match lated from several species, suggesting that the protein being to the Drosophila splicing regulators Tra (Boggs et family is evolutionarily well conserved from Drosophila al. 1987), Tra2 (Amrein et al. 1988; Goralski et al. 1989), to humans (see Fig. 1, legend). Further, the proteins are and su (w^) (Chou et al. 1987). A considerable number of all recognized by a previously described monoclonal an­ proteins containing RS domains or related motifs has tibody (mAbl04; Roth et al. 1990) that targets a phos- since been identified, and the presence of an RS-like re­ phorylated epitope residing within the RS domain. Par­ gion continues to be diagnostic of a protein involved in tial amino acid sequence revealed that the proteins all splicing. showed sequence similarity, and two, of —30 kD, were SC35 was the next SR protein to be identified. It was found to be identical to ASF/SF2 and SC35. The purified detected first with a monoclonal antibody raised against proteins also all have the ability to function as essential purified spliceosomes and was shown to be necessary for splicing factors. Indeed, a recombinant form of one from splicing and spliceosome assembly in vitro (Fu and Ma­ Drosophila, SRp55, was found to function interchange­ niatis 1990). Isolation of cDNAs encoding SC35 revealed ably with human ASF/SF2 in in vitro assays (Mayeda et a primary structure very similar to that of ASF/SF2: an al. 1992), adding to the view that the function of these amino-terminal RBD and a carboxy-terminal RS domain proteins, though conserved throughout metazoan spe­ (Fu and Maniatis 1992a). The two proteins were also cies, might be redundant. shown to function interchangeably in in vitro assays: The partial amino acid sequence data of Zahler et al. Each by itself could activate splicing of test pre-mRNAs (1992) along with cDNA sequences of the SR proteins in depleted extracts, and each could induce the same provide a structural basis for the defining of an SR pro­ switch in selection of alternative splice sites in model tein (see Fig.